Where is the math?

[arthurwolf]

Hey.

Where do I find the actual PID tuning math?

I've gone over the files, and I can't find an obvious place that looks math-intensive.

Is it in pid/v1.0.6/pid_tuner.js ? Is it pre-compiled as wasm ? Is it just not available and accessed through calls to a remote API ??

Thanks a lot in advance. Best regards.

[pidtuner]

Hi, the implemented algorithm is the one from the following paper:

https://folk.ntnu.no/skoge/publications/2001/tuningpaper_reno/tuningpaper_06nov01.pdf

It is indeed pre-compiled for performance purposes, all maths are computed in the browser.

there is a remote API when you call the Save button, but that just saves a JSON into a mongo server containing the project data, you can see the data sent using your browser's dev tools.

[arthurwolf]

Is it your intention to hide the math/code?

If not, it would be very much appreciated if you could publish the un-compiled source code for the actual PID math/tuning along with the compiled version in the same places/folder, it would be helpful to people trying to better understand this, and considering implementing something similar (in my case, maybe embedded in the Smoothieware controller/board/firmware ).

Thanks!

[pidtuner]

The intention is to make it work in a reasonable amount of time, because the math routines are very complicated and linear algebra heavy, and is all implemented in C++ using armadillo with BLAS and LAPACK routines. I don't think I could implement this in javascript directly in feasible time.

The C++ source code is very messy at the moment, I will put in my TODO list to make it publishable, but it will be low in my priority list.

[arthurwolf]

That's very good news, thank you very much for planning to do it, very much appreciated.

On Wed, May 26, 2021 at 5:50 PM PidTuner @.***> wrote:

The intention is to make it work in a reasonable amount of time, because the math routines are very complicated and linear algebra heavy, and is all implemented in C++ using armadillo http://arma.sourceforge.net/ with BLAS and LAPACK routines. I don't think I could implement this in javascript directly in feasible time.

The C++ source code is very messy at the moment, I will put in my TODO list to make it publishable, but it will be low in my priority list.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/pidtuner/pidtuner.github.io/issues/10#issuecomment-848889426, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAA2SFMOKTJZ4YUL7R57UUTTPUKDFANCNFSM45SGD3XQ .

--

勇気とユーモア

[pidtuner]

Apart from the tuning, the identification algorithm uses this solution. Hopefully this will help you in the meanwhile.

[aditya-gumparthi]

I would love to see the c++ code, in any form. Anything I/we can do to get it out there faster?

[pidtuner]

Hire me full time to maintain this project :slightly_smiling_face: This issue is low priority, the few hours I have on my free time will go to make this useful to the most people. At the moment I will focus on the drone people. Hope you understand.

[Meach]

Hey! I would also be intested to see the c++ code if possible. I have been studying the paper you posted but cannot understand exactly how are the integrator with lag variables calculated. Same with the PID gain values on the following step.

[pidtuner]

With kids and a full time job is impossible for me to give some love to this project, I don't even remember where is the repo with the sources. At this rate you guys are going to have to wait for my retirement. Sorry.